7 Appendix

GENSMI Algorithm

Input: SMILES, Output: indefinite number of SMILES

molecule <- dt_smilin(seed) // interpret a SMILES string

MakeEdge(molecule) // form cycles recursively

MakeNode(molecule) // add an atom and form cycles recursively

dt_dealloc(molecule) // remove object

Figure 1: Pseudo-Code for generating SMILES. The seed can be an atom or a fragment.
MakeNode Routine

Input: molecule, Output: indefinite number of SMILES

NotDone() // check limits on number of SMILES, memory use, time elapsed

setA <- SequenceOfAtoms(molecule, TYP_ATOM) // like dt_stream()

setA <- DeleteRedundancy(setA) // uses dt_symclass() for atom uniqueness

atomA <- dt_next(atomsA) // loop of set A

OkToConnect(atomA) // check valence and ring membership

i <- 1 to atom types // loop over elements - C, N, O, etc.

i <- 1 to bond types // loop over bond orders - single, double, etc.

Index(atomA) > PrevIndexA(molecule) // avoid redundancy between recursion levels

and AtomComposition(molecule, atom typei) // check distribution

and ApproxMass(molecule, 2*atom typei + atom valencei - 2*bond typei) // check maximum

molecule <- Extend(atomA, bond typei) // uses dt_addbond()

Branching(molecule) // number of bonds per non-terminal atom

and Complexity(molecule) // SCMC - Allu and Oprea, CUP IV

and Cyclization(molecule) // number of cycles per ring atom

and ChemConstraints(molecule) // uses dt_match() for SMARTS patterns

and UniqueSmiles(molecule) // uses dt_cansmiles() for graph isomorphism

Similarity(molecule) // uses dt_fp_tanimoto(), dt_fp_tversky(), and dt_fp_euclid()

Output(molecule) // write SMILES and stereoisomers, uses du_chiralify()

MakeEdge(molecule) // form cycles recursively

MakeNode(molecule) // add an atom and form cycles recursively

molecule <- UnExtend(atomA, bond typei) // uses dt_dealloc()

Figure 2: Pseudo-Code for adding an atom. Input requires a molecule. Passage of all constraints results in output of a SMILES string and recursion, otherwise the routine returns.

MakeEdge Routine

Input: molecule,Output: indefinite number of SMILES

NotDone() // check limits on number of SMILES, memory use, time elapsed

setA <- SequenceOfAtoms(molecule, TYP_ATOM) // like dt_stream()

setA <- DeleteRedundancy(setA) // uses dt_symclass() for atom uniqueness

setZ <- SequenceOfAtoms(molecule, TYP_ATOM)

atomA <- dt_next(atomsA) // loop of set A

OkToConnect(atomA) // check valence and ring membership

setB <- RingSet(atomA, setZ) // check A-B patch length

atomB <- dt_next(setB) // loop over set B

Index(atomA) < Index(atomB) // avoid redundancy of A,B and B,A

and NotConnected(atomA, atomB) // check for existing bond

and OkToConnect(atomB) // check valence and ring membership

and RingStrain(atomA, atomB) // avoid pro-spiral and bridging aromatic systems

i <- 1 to bond types // loop over bond orders - single, double, etc.

Index(atomA) > PrevIndexA(molecule) // avoid redundancy between recursion levels

and BondComposition(molecule, bond typei) // check distribution

and ApproxMass(molecule, 2*bond typei) // check minimum

molecule <- Cyclize(atomA, atomB, bond typei) // uses dt_addbond()

Branching(molecule) // number of bonds per non-terminal atom

and Complexity(molecule) // SCMC - Allu and Oprea, CUP IV

and Cyclization(molecule) // number of cycles per ring atom

and ChemConstraints(molecule) // uses dt_match() for SMARTS patterns

and UniqueSmiles(molecule) // uses dt_cansmiles() for graph isomorphism

Similarity(molecule) // uses dt_fp_tanimoto(), dt_fp_tversky(), and dt_fp_euclid()

Output(molecule) // write SMILES and stereoisomers, uses du_chiralify()

MakeEdge(molecule) // form cycles recursively

molecule <- UnCyclize(atomA, atomB, bond typei) // uses dt_dealloc()

Figure 3: Pseudo-Code for forming a cycle. Input requires a molecule. Passage of all constraints results in output of a SMILES and recursion, otherwise the routine returns.